Previous work by Niko Fagerström

maskmill.com

AI background removal for Maskmill orders and image delivery

Maskmill needed AI background removal that works as part of each customer order. I designed and built the connection between image processing, payments, file storage and production.

The workflow I built automatically produces PNG, JPG and mask files and preserves the source material for corrections. Customers receive the finished files through the same service, and production staff can make human corrections when needed.

10s
worker polling cycle
50MP
image processing class
3
output asset types
1
complete order process
Traditional approach

A separate AI tool with manual handoffs.

  • Customer details, payments and images in separate places
  • Manual retrieval of finished images and masks
  • Correction orders require recreating the original work package
  • Customer has no real-time processing feedback
What I built

Background removal and image delivery belong to the same order.

  • After payment, images move automatically into the processing queue
  • Background worker generates the mask, transparent PNG, color-backed JPG, and preview
  • The background mask is saved for corrections
  • Customers can see how processing is progressing

Scope: Order form, upload validation, post-payment queueing, background processing, image compositing, cloud storage, previews, and correction workflow.

.NET Blazor / Razor EF Core Background Service SignalR ImageMagick Cloud Blob Storage External AI Image API
Pipeline

How background removal runs in production

Images are processed from a queue, with progress tracked for each file. Failed jobs can be handled separately. Finished images become available for the customer to download.

  1. 1

    Order and settings

    The customer uploads images and selects crop and background color options. Those settings are stored directly on the order.

  2. 2

    Validation

    Oversized or unsupported images are converted to fit the processing limits before AI handling starts.

  3. 3

    Queueing

    After payment, files are marked as waiting for processing. A background service picks up the queue automatically.

  4. 4

    Mask and finished image

    An external AI service returns the image mask. The system combines it with the original image and creates the finished customer files.

  5. 5

    Delivery and corrections

    Finished images are saved for download. The background mask is kept separately so an editor has the right files for any corrections.

Architecture

How does image processing keep running?

01

Progress is tracked for every image

Image processing continues even if the user closes the browser. Each file shows whether it is waiting, ready or needs attention after an error.

02

From AI output to a finished image

AI separates the subject from the background. The system then creates an image with a transparent or chosen background, along with a lightweight preview.

03

Manual corrections when needed

If an image needs a manual correction, the original image and background mask are ready. The editor does not need to start again.

What this work demonstrates

I connected AI background removal with Maskmill orders, payments and file handling. I also built a way to correct images manually when the AI result needs more work.

Want to add AI to your service? Tell me where it could help.

Contact me → All work samples